TDTS 16 Round-the-clock, on-line and cryogen-free monitoring of hydrocarbons from acetylene to trimethylbenzene in ambient air

Transcription

1 Application TDTS 16 Round-the-clock, on-line and cryogen-free monitoring of hydrocarbons from acetylene to trimethylbenzene in ambient air Summary This Application Note describes validation of a thermal desorption system for on-line monitoring of a complex mix of volatile and very volatile hydrocarbons ( ozone precursors ) in air. Background The presence of volatile hydrocarbons in urban atmospheres is believed to contribute to the formation of ground-level ozone, one of the main constituents of urban smog. The compounds of interest range in volatility from acetylene (ethyne) to trimethylbenzene, and are generally referred to as 'ozone precursors'. Vehicle emissions are thought to be the main source of these compounds. Recent European and US regulations 1,2 require round-the-clock monitoring of target species in all major urban centers to establish and monitor the link between periods of high traffic density and high pollution levels (key compounds include benzene, toluene, xylene and buta-1,3-diene). Continuous real-time monitoring provides an insight into emission episodes from local industry and can be used to monitor the effect of weather conditions such as wind direction, precipitation and temperature inversion. Introduction The UNITY 2 Air Server is a cost-effective system for round-the-clock speciated measurement of multiple trace-level volatile organic compounds (VOCs) in air or pure gases. It combines automated, controlled-flow sampling with cryogen-free concentration technology. The system connects to standard GC and GC/MS technology, and is designed for unattended operation in remote field locations. In summary, sample air is pulled directly onto an electrically-cooled, sorbent-packed focusing trap at a controlled flow rate. No liquid cryogen is required. An optional membrane dryer selectively eliminates water and other low-molecular-weight polar components, reducing the risk of interference from unknown species and facilitating the use of FID rather than MS detection. Sampling flows are controlled by an electronic mass flow controller and pump, located downstream of the focusing trap to avoid contamination. All sampling parameters are selected by the user and monitored by system software as an integral part of the analytical method. After sample collection, the flow path is purged with carrier gas to prevent carryover and eliminate oxygen from the focusing trap. The trap then heats rapidly, at rates approaching 100 C/min, to inject retained analytes into the capillary column as a highly concentrated band of vapour. This transfer may be performed splitless for maximum sensitivity. Once the trap has desorbed, it re-cools, reequilibrates to the trapping temperature and begins collection of the next sample while analysis of the previous sample is ongoing. The UNITY 2 Air Server also offers automatic interchange between a minimum of three sample channels (typically sample, reference and blank) for remote system calibration/validation as per user requirements. An 8-channel version is also available. This Application Note describes validation of the system for on-line monitoring of the 27 ozone precursors specified by European regulators, with the addition of 1,2,4-trimethylbenzene, 1,3,5-trimethylbenzene and isoprene (present in the gas standard employed). Table A1 (see Appendix) provides the full list of 30 compounds. Key factors taken into account when developing this method included the need for a system that could operate round-the-clock unattended, so cryogen-free operation was essential. The system also needed to allow hourly sample collection, with as much of the hour as possible dedicated to sampling. Finally, detection limits needed to be below 0.5 ppb (ideally 0.1 ppb).

2 Page 2 Defining the ideal focusing trap The design and packing of the trap is crucial for this application for many reasons. The narrow-bore design of the UNITY 2 trap (Figure 1) allows true splitless operation, allowing transfer of the whole sample to the GC system, giving increased sensitivity for lowconcentration compounds. The narrow-bore design also gives the best possible peak shape for early-eluting compounds (Figure 2). Quartz wool Weak sorbent Strong sorbent Strongest sorbent Retaining spring Figure 1: Design of the focusing trap. Note the narrow-bore outlet and the long sorbent bed. 1 Ethane 2 Ethene 3 Acetylene 4 Propane Propene 6 2-Methylpropane 7 Butane 8 trans-but-2-ene Figure 2: Chromatogram showing a splitless injection of 500 ml of the calibration standard. 6 8 Experimental (a) Breakthrough tests TD: Prepurge time: 1 min Sampling rate: 25 ml/min Sampling time: Various Cold trap: U-T17O3P-2S (ozone precursor) Cold trap low: 30 C Cold trap high: 325 C Cold trap hold: 5 min Flow path temp.: 80 C GC: Column: Agilent J&W GS-GasPro, 30 m 0.32 mm GC program: 40 C (hold 6 min), then 20 C/min to 200 C (hold 1 min) (b) Two-column Deans switch GC system TD: Prepurge time: 2 min Sampling rate: 25 ml/min Sampling time: Various Cold trap: U-T17O3P-2S (ozone precursor) Cold trap low: 30 C Cold trap high: 325 C Cold trap hold: 5 min Flow path temp.: 80 C GC/FID: Inlet pressure: 44.9 psi Midpoint pressure: psi GC oven program: 30 C (hold 12 min), then 5 C/min to 170 C, then 15 C/min to 200 C Switching time: 17.5 min FID temp.: 250 C Air flow: 400 ml/min H 2 flow: 40 ml/min The sorbents chosen are also crucial, as they permit increased retention of very volatile compounds, thereby increasing the sample volume taken, and allowing lower levels to be detected.

3 Page 3 Results and discussion (a) Breakthrough tests All the C 2 hydrocarbons are highly volatile species. Acetylene (ethyne) has a boiling point of 84 C and is the most difficult to trap, and to quantitatively retain it without liquid cryogen requires careful selection of the cold trap sorbent(s) and focusing temperature. Experiments to determine the limit of quantitative retention (the breakthrough volume) were performed on an ozone precursor cold trap using the ppb-level certified gas standard generated by the National Physical Laboratory (Teddington, UK). Steadily increasing volumes of the gas standard were introduced, and a graph of detector response against sampled volume plotted. Figure 3 is a plot of the peak area against the sampled volume of calibration gas for acetylene, and shows that even at 1.5 L of sampled gas, there is no deviation from linearity, i.e. there is negligible breakthrough. FID 1 Sample in FID 2 Column 3 ~0.5 m, 0.1 mm uncoated deactivated fused silica Column 1 60 m, 0.25 mm, 1 µm dimethylpolysiloxane Column 2 50 m, 0.32 mm, 8 µm alumina PLOT (Na 2 SO 4 wash) Switching point Pressure controller Figure 4: Schematic representation of the two-column ozone precursor analytical system. Acetylene peak area R 2 = Sample volume (ml) Figure 3: Plot of peak area against volume sampled for acetylene. Image courtesy of Ecole des Mines de Douai. (b) Two-column Deans switch GC system The column used in all previous tests was a 30 m 0.32 mm i.d. Agilent J&W GS-GasPro. Unfortunately, this column did not allow full resolution of the C 4 and C 5 components in the standard mixture. For better separation of these components, it was necessary to use a two-column system with a Deans switch (schematically illustrated in Figure 4). All the effluent from the focusing trap of the desorber was directed into the dimethylpolysiloxane column at 40 C. The C 2 C 5 compounds are unresolved on this column at this temperature, and are passed, via the Deans switch, into the alumina PLOT column as they elute from the primary column. For the first 17.5 minutes of the run, the Deans switch between the primary and secondary columns works in this direction. At this point, all the compounds eluting from the primary column are C 6 and higher, and are well-resolved. The Deans switch is thus activated, directing effluent from the primary column to the deactivated, uncoated fused silica link and to the other FID. From this point on, two chromatograms are produced in parallel; the C 6+ compounds from the dimethylpolysiloxane column (labelled FID 1), and the C 2 C 6 compounds from the alumina PLOT column (labelled FID 2). Figure 5 shows a plot for the calibration gas run through the two-column system, and Figure 6 shows the equivalent plot for the 56-component US EPA ozone precursor mix.

4 Page 4 1 Ethane 2 Ethene 3 Propane 4 Propene 5 2-Methylpropane 6 n-butane 7 Acetylene 8 trans-but-2-ene 9 But-1-ene 10 cis-but-2-ene 11 2-Methylbutane 12 Pentane 13 Buta-1,3-diene 14 trans-pent-2-ene 15 Pent-1-ene 16 2-Methylpentane 17 Isoprene 18 n-hexane 19 Benzene 25 m- and p-xylene 20 2,2,4-Trimethylpentane n-heptane 27 o-xylene 22 Toluene 28 1,3,5-Trimethylbenzene 23 Octane 29 1,2,4-Trimethylbenzene 24 Ethylbenzene 30 1,2,3-Trimethylbenzene Alumina PLOT column (FID 2) Dimethylpolysiloxane column (FID 1) Figure 5: Chromatograms for the ppb-level 30-component calibration standard (generated by NPL) on the two-column analytical system. All components had a concentration in the range ppb, with uncertainties for each of ±0.08 ppb. 1 Ethane 2 Ethene 3 Propane 4 Propene 5 2-Methylpropane 6 n-butane 7 Acetylene 8 trans-but-2-ene 9 But-1-ene 10 cis-but-2-ene 11 Cyclopentane 12 2-Methylbutane 13 Pentane 14 trans-pent-2-ene 15 Pent-1-ene 16 cis-pent-2-ene 17 2,2-Dimethylbutane 18 2,3-Dimethylbutane 19 2-Methylpentane 20 3-Methylpentane 21 Isoprene 22 2-Methylpent-1-ene 23 n-hexane 24 Methylcyclopentane 25 2,4-Dimethylpentane 26 Benzene 27 Cyclohexane 28 2-Methylhexane 29 2,3-Dimethylpentane 30 3-Methylhexane 31 2,2,4-Trimethylpentane 32 Heptane 33 Methylcyclohexane 34 2,3,4-Trimethylpentane 35 Toluene 36 2-Methylheptane 37 3-Methylheptane 38 Octane 39 Ethylbenzene 40 m- and p-xylene 41 Styrene 42 o-xylene 43 Nonane 44 Isopropylbenzene 45 Propylbenzene 46 m-ethyltoluene 47 p-ethyltoluene 48 1,3,5-Trimethylbenzene 49 o-ethyltoluene 50 1,2,4-Trimethylbenzene Decane 52 1,2,3-Trimethylbenzene 53 m-diethylbenzene 54 p-diethylbenzene 55 Undecane 56 Dodecane Figure 6: Chromatograms for the ppb-level 56-component calibration standard (from the US EPA) on the two-column analytical system. Data courtesy of ITC, Korea

5 Page 5 Reproducibility of retention time Due to the nature of the two-column set-up, the stability of the retention time can be related to the pressure balance in the system. With the added capability of electronic carrier gas control (ECC) on the UNITY 2 Air Server, stable retention times on both columns are achieved. Tables A1 and A2 (see Appendix) show the relative standard deviations for the compounds in the 30- component standard; those on the dimethylpolysiloxane column (C 6+ compounds) were close to 0.01%, and those in the alumina PLOT column (C 2 C 6 compounds) were in the range %. Reliability study multipoint calibrations In order for a system to be reliable, stringent tests must be carried out. Figure 7 shows five repeat analyses of 500 ml of the ozone precursor calibration standard on the two-column system. Alumina PLOT column (FID 2) Dimethylpolysiloxane column (FID 1) Figure 7: Repeat analysis of a splitless gas injection of a 500 ml sample of gas standard.

6 TDTS 16 Page 6 As can be seen, the repeat analyses appear identical, but to confirm this numerically, a 5-point calibration curve was generated, and five replicates at each level were analysed (see Table A3 in the Appendix). Nearly all the relative standard deviations were below 1%, demonstrating the reliability of the method. Each of the repeat calibrations was then plotted and a linear relationship derived; the corresponding R 2 values are provided in Table A4 (see Appendix). R 2 values of 0.99 or above were seen for more than 98% of the data, confirming the reliability of the calibration. French norm NF T specifies a method for testing calibration data and systems in accordance with a stringent goodness-of-fit test. We used this F-test method to compare a theoretical calibration against the five 5-point experimental calibrations, assuming the homogeneity of the variances. The critical value (F crit ) corresponds to the value that F must exceed for this assumption to be incorrect, and in this case it is at a risk level (α) of 1%. Table 1 shows the F-values for each of the compounds, and they are all less than F crit, so the calibration is acceptable. Detection/quantitation limits Signal-to-noise ratios for the 500 ml volume of standard using the two-column system are in the order of 250:1 for the light hydrocarbons at ~4 ppb, and 350:1 for compounds from n-butane upwards at ~4 ppb. Assuming a minimum detection signal of 3:1, approximate detection/quantitation limits for the 500 ml air samples were calculated as follows: C 2 C 3 hydrocarbons: Detection: 0.05 ppb; quantification: 0.1 ppb C 4+ hydrocarbons: Detection: 0.03 ppb; quantification: 0.06 ppb Compound Calculated F-value Ethane Ethene Propane Propene Methylpropane n-butane Acetylene trans-but-2-ene But-1-ene cis-but-2-ene Methylbutane Pentane Buta-1,3-diene trans-pent-2-ene Pent-1-ene Methylpentane Isoprene n-hexane n-heptane Benzene ,2,4-Trimethylpentane Octane Toluene Ethylbenzene m-xylene p-xylene o-xylene ,3,5-Trimethylbenzene ,2,4-Trimethylbenzene ,2,3-Trimethylbenzene Table 1: F-test values.

7 Page Concentration (ppb) Benzene Toluene Propane 2-Methylpropane n-butane Acetylene 2-Methylbutane Buta-1,3-diene :00 20:00 00:00 04:00 08:00 12:00 Time (GMT) 16:00 Figure 8: Diurnal profile of selected hydrocarbons, showing distinct differences between traffic-free and traffic-heavy periods. Suburban air monitoring Analyser performance for unattended field operation was evaluated in a forested semi-rural area, with high volumes of traffic during the working week and very few cars at weekends. Figure 8 shows a diurnal profile of various compounds from 16:40 on a Sunday through to a Monday evening the difference between the two periods is distinct. Most of the hydrocarbons were at low levels during the traffic-free periods, whilst rapid increases were seen during peak commuting hours. These data were collected at the end of the trial, and show that good analytical performance was maintained. Importantly, no user intervention was required during the trial, while detection limits, even in this relatively clean environment, were also satisfactory. Conclusions A reliable, semi-continuous, cryogen-free sampling and analysis system has been demonstrated for this challenging and complex mix of volatile and very volatile hydrocarbons, confirming Markes UNITY 2 Air Server system as an excellent choice for those wishing to quantify ozone precursors in ambient air. References 1. Council Directive 96/62/EC of 27 September 1996 on ambient air quality assessment and management, Council of the European Parliament, 1996; and its daughter directive, Directive 2000/69/EC of the European Parliament and of the Council of 16 November 2000 relating to limit values for benzene and carbon monoxide in ambient air, Council of the European Parliament, See for more details. 2. Provisions for attainment and maintenance of national ambient air quality standards (US Clean Air Act Amendments of 1990). See for more details. 3. Revision de la norme NF T90-210: Qualite de l eau. Protocole d évaluation initiale des performances d une méthode dans un laboratoire, Association Francaise de Normalisation, Available from gne.aspx?&nivctx=nelznelz1a10a101a107&ts= &CLE_ART=FA Applications were performed under the stated analytical conditions. Operation under different conditions, or with incompatible sample matrices, may impact the performance shown.

8 TDTS 16 Page 8 Appendix Compound Retention time RSD (%) (n = 15) Benzene ,2,4-Trimethylpentane n-heptane Toluene Octane Ethylbenzene m- and p-xylene o-xylene ,3,5-Trimethylbenzene ,2,4-Trimethylbenzene ,2,3-Trimethylbenzene Table A1: Retention time stability on the dimethylpolysiloxane column (C 6+ compounds). Compound Retention time RSD (%) (n = 15) Ethane Ethene Propane Propene Methylpropane n-butane Acetylene trans-but-2-ene But-1-ene cis-but-2-ene Methylbutane Pentane Buta-1,3-diene trans-pent-2-ene Pent-1-ene Methylpentane Isoprene n-hexane Table A2: Retention time stability on the alumina PLOT column (C 2 C 6 compounds).

9 TDTS 16 Page 9 Response RSD (%) (n = 5) Compound 250 ml 375 ml 500 ml 625 ml 750 ml Ethane Ethene Propane Propene Methylpropane n-butane Acetylene trans-but-2-ene But-1-ene cis-but-2-ene Methylbutane Pentane Buta-1,3-diene trans-pent-2-ene Pent-1-ene Methylpentane Isoprene n-hexane Benzene ,2,4-Trimethylpentane n-heptane Toluene Octane Ethylbenzene m- and p-xylene o-xylene ,3,5-Trimethylbenzene ,2,4-Trimethylbenzene ,2,3-Trimethylbenzene Table A3: Reproducibility of the 30-component gas standard for five sample volumes.

10 Page 10 Compound Calibration 1 Calibration 2 Calibration 3 Calibration 4 Calibration 5 Ethane Ethene Propane Propene Methylpropane n-butane Acetylene trans-but-2-ene But-1-ene cis-but-2-ene Methylbutane Pentane Buta-1,3-diene trans-pent-2-ene Pent-1-ene Methylpentane Isoprene n-hexane Benzene ,2,4-Trimethylpentane n-heptane Toluene Octane Ethylbenzene m- and p-xylene o-xylene ,3,5-Trimethylbenzene ,2,4-Trimethylbenzene ,2,3-Trimethylbenzene Table A4. R 2 values for five repeats of the five-point calibration. Version 12 June 2012